Received February 17, 2022, accepted March 11, 2022, date of publication March 17, 2022, date of current

version March 25, 2022.

Digital Object Identifier 10.1109/ACCESS.2022.3160484

Solar Power Forecasting Using Deep

Learning Techniques
MEFTAH ELSARAITI , (Member, IEEE), AND ADEL MERABET , (Senior Member, IEEE)
Division of Engineering, Saint Mary’s University, Halifax, NS B3H 3C3, Canada
Corresponding author: Meftah Elsaraiti (meftah.elsaraiti@smu.ca)

ABSTRACT The recent rapid and sudden growth of solar photovoltaic (PV) technology presents a future
challenge for the electricity sector agents responsible for the coordination and distribution of electricity given
the direct dependence of this type of technology on climatic and meteorological conditions. Therefore, the
development of models that allow reliable future prediction, in the short term, of solar PV generation will
be of paramount importance, in order to maintain a balanced and comprehensive operation. This article
discusses a method for predicting the generated power, in the short term, of photovoltaic power plants,
by means of deep learning techniques. To fulfill the above, a deep learning technique based on the Long
Short Term Memory (LSTM) algorithm is evaluated with respect to its ability to forecast solar power data.
An evaluation of the performance of the LSTM network has been conducted and compared it with the Multi-
layer Perceptron (MLP) network using: Mean Absolute Error (MAE), Mean Absolute Percentage Error
(MAPE), Root Mean Squared Error (RMSE) and Coefficient of Determination (R2 ). The prediction result
shows that the LSTM network gives the best results for each category of days. Thus, it provides reliable
information that enables more efficient operation of photovoltaic power plants in the future. The binomial
formed by the concepts of deep learning and energy efficiency seems to have a promising future, especially
regarding promoting energy sustainability, decarburization, and the digitization of the electricity sector.

INDEX TERMS Prediction, deep learning, solar power, time series.

I. INTRODUCTION nonlinear behaviors. In recent decades, the application of

Renewable energy, especially solar PV, will gain prominence
as a major source of energy in the future. But as their share
of the energy mix grows, ensuring the safety, reliability,
and profitability of power generation assets will be a top
priority. Therefore, the successful integration of solar energy
into the electrical grid requires an accurate prediction of
the power generated by photovoltaic panels. Speaking of
solar energy in particular, its unexpected behavior brings
with it a series of problems when generating energy, such
as voltage variations, power factor details, and stability. For
this reason, these new tools are constantly being created
that contribute to the prediction of future events, with the
aim of reducing errors in predictions [1]. Auto-Regressive
Integrated Moving Averages (ARIMA) models have proven
to be extremely useful for the short-term prediction of high-
frequency time series. In contrast to ARIMA models and sta-
tistical methods, artificial neural networks are more powerful,
especially in representing complex relationships that exhibit
The associate editor coordinating the review of this manuscript and
approving it for publication was Giambattista Gruosso
artificial neural networks (ANNs) in time series prediction
has grown due to the ideal characteristics offered by ANNs
for working with nonlinear models. Likewise, the develop-
ment of applications that facilitate work when carrying out
simulations with ANN continues to increase. In [2], artificial
neural networks are highlighted as one of the prediction
methods for time series thanks to their great adaptability and
capacity to solve nonlinear and complex problems. In recent
years, as a result of the research on artificial intelligence,
deep learning based on ANN has come to light to become
very popular due to its capability to accelerate the solution
of some difficult computer problems [3]. While multi-layer
perceptron (MLP)-type ANNs can be used to model complex
relationships, they are incapable of assimilating the long- and
short-term dependencies present in historical data [4]. These
dependencies refer to the ability of an ANN to identify and
remember behavior patterns from the distant past and the near
past, respectively. An ANN is needed to make predictions
of sequential data behavior [5]. As an attempt to address
this problem, the first Recurrent Neural Networks (RNN)
. emerged in the 1980s, where the term ‘‘recurrent’’ refers to

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
31692 VOLUME 10, 2022
M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques
the characteristic of these networks to have internal feed-
back loops. However, these networks presented a great dis-
advantage, which is known in the literature as the vanishing
gradient problem. This problem refers to the difficulty of
training these networks with methods based on gradients
or backpropagation algorithms. This is due to the fact that
when this type of method calculates the weight adjustment
based on the chain rule, there are common multiples between
0 and 1 that are multiplied by n times. Numbers less than
1multiplying by n times, an exponential decrease in the error
signal is generated, which significantly affected the training
of these RNNs [6]. Due to the above drawbacks, Long Short
Term Memory (LSTM)-type RNNs emerged in 1997 as a
solution due to their memory units in the network cells [7].
The prediction of solar radiation is a fundamental key to
increasing the supply of electrical energy generated from this
medium to the distribution networks. In the energy market,
when an electric energy producer does not comply with the
programmed offer, they are penalized with a proportional
relationship between the energy actually produced and what
is stated in the contract. The integration of these renewable
energy sources intensifies the complexity of managing the
grid and the ongoing balance between electricity consump-
tion and production due to its unpredictable and intermittent
nature. Several tools and methodologies have been developed
for the prediction of solar energy at different horizons. In this
study, the task of predicting the photovoltaic power for the
coming days in short time intervals (30 minutes) from data
previously recorded during one year was considered using
Long Short Term Memory (LSTM) Recurrent Neural Net-
works (RNN). This is a type of (ANN) that estimates the next
data from the data received in the past and the current input
data and compares it using a multi-layer perceptron (MLP)
neural network, which is a neural network that has limita-
tions in performance because it learns only using the one-
to-one relationship between input and output and does not
consider the characteristics of time series data. The results
of each scheme are evaluated using: Mean Absolute Error
(MAE), Mean Absolute Percentage Error (MAPE), Root
Mean Squared Error (RMSE), and Coefficient of Determi-
nation (r 2 ). A preliminary study of the data was performed
to find patterns and apply certain corrections, such as elim-
inating missing data by averaging the previous 60 minutes’
values (average 12 previous values). The remainder of the
paper is organized as follows: Section 2 provides a review of
previous works on the topic, Section 3 provides an explana-
tion of the various deep learning techniques based on LSTM
and MLP, Section 4 presents and discusses the results, and
Section 5 concludes the paper with a discussion of future
suggestions.
II. LITERATURE REVIEW
Artificial Neural Networks (ANNs) are of great importance
in contemporary research on the mathematical modeling of
systems based on renewable energies. It reported higher
solar forecast results than those obtained using traditional
VOLUME 10, 2022
statistical methods [8]. In the presence of noisy data, the
LSTM model provides a more accurate prediction compared
to the ARIMA model [9]. The use of deep learning techniques
has allowed excellent results in classification and regression
problems to be obtained. This is due to its automatic adjust-
ment of internal parameters by means of supervised learning
algorithms [10]. Research has focused on describing vari-
ous artificial intelligence (AI) techniques used in renewable
energy for their prediction and optimization [11]. Existing
artificial intelligence tools have been used for the model-
ing and simulation processes of solar energy systems [12].
This study examines the importance and benefits of artificial
neural networks in predicting environmental variables and
estimating some unconventional energy systems, as well as
describing the main methodologies and neural models used
to date [13]. Recently, LSTM has become one of the most
widely used artificial neural networks for short-term wind
speed predictions [14], [10]. The LSTM has been proposed
to make forecasts of wind speed on 24-hour wind farms.
This work compares the results obtained by LSTM with
MLP, deep versions of MLP, and classical methods. LSTM
showed better efficiency prediction results than MLP and
classical methods [15]. Neural networks-based deep learning
are widely used in solar radiation applications. Evaluating
the performance of the multilayer perceptron (MLP) and the
empowered decision trees by combining them with linear
regression for the estimation of solar energy; the results show
that the MLP model had a better performance according to
the coefficient of determination indicators (R2 ) which was
97.7% and the RMSE, which was 0.033 [16]. A long-term
prediction approach was used with the MLP network and
iterative strategy; however, their results show that the support
vector machine (SVM) model with a straightforward strategy
yields better results [17]. LSTMs have the characteristic of
being able to process large amounts of data and have the
ability to generalize; that is, they adapt to unknown data,
which makes them get better results than SVM-based mod-
els [18]. The Long-Term Memory Recurrent Neural Network
(LSTM-RNN) has been proposed to accurately predict the
output power of solar PV systems using hourly data sets
for a year and has compared the proposed method results
with multiple linear regression (MLR), bagged regression
trees (BRT), and neural networks (NNs) methods for pho-
toelectric prediction. The LSTM networks prove less pre-
dictive error compared to other methods [19]. Long short
term memory (LSTM) neural networks were used to predict
solar power panels and wind power in the medium and long
term. The errors obtained were lower than those of the per-
sistence model and the SVM model [20]. As an alternative
to the conventional approaches, ANNs have been success-
fully applied to estimating solar irradiance. The multilayer
perceptron structure (MLP) is the most common neural net-
work [21]. An MLP method has been proposed to predict
solar radiation for the next 24 hours using current data of
mean solar radiation and air temperature for a region in
Italy [22].
31693
 M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques

III. METHODOLOGY
In this section, the sources of information and data processing
methods are described, and the building of deep learning-
based neural networks to predict solar radiation is presented.

A. SOURCES OF INFORMATION
In this study, the MATLAB software (R2019b) was used
for the training process of the LSTM, which is an advanced
architecture for RNN to predict the values of future time
steps of a sequence. The sequence regression network was
trained to the LSTM sequence, where responses are training
sequences with changing values in one time step. That is,
FIGURE 1. Shows the original photoelectric data.
for each time step of the input sequence, the LSTM network
learns to predict the value of the next time step. To evaluate
the effectiveness of the proposed method, a case study was
performed using a data set obtained from Nova Scotia Com-
munity College in Halifax that includes one-year data (from
January 1, 2017, to December 31, 2017) for Halifax, located
in Nova Scotia, Canada. For each day, data was selected only
during daylight hours, from 8 am to 5 pm. The original photo-
electric data were collected at 5-minute intervals and included
365 × 120 = 43800 measurements. Each missing value is
FIGURE 2. Schematic representation of neurons.
processed by averaging the previous 60 min values. Data is
summarized at 30-minute intervals as the task is to make a highly connected to each other, to work together to solve a
forecast for each half-hour for the next day. Thus, we have problem. Deep learning techniques can be used in forecasting
20 values in one day and 20 × 365 × 1 = 7300 values in one and identifying characteristics of each one; among them:
year. The data is then normalized to the interval [0-1]. Upon stages implemented for the approximation, the accuracy with
completion of the forecasting process, the results obtained for which the power generation is approximated, the convergence
the proposed models are compared with the actual value. The time and the uncertainty associated with the forecast. In the
fact that the observation recording period is 5 minutes. Thus, case of neural networks, studies have demonstrated the ability
within one hour, you will have 12 observations. In turn, 188 of this technique to accurately determine the time series of
(12 × 24) values. The number of past values of the series data.
that are considered to forecast the target vector is independent
of the size of the target vector and will depend in each case C. MULTILAYER PERCEPTRONS (MLP)
on the algorithm used and the nature of the problem. In this
The Multilayer Perceptron (MLP) network is the most popu-
strategy, a single execution of the algorithm, with the passed
lar ANN architecture used in solving scientific problems [26],
data that you need, results in the forecast with the necessary
due to its demonstrated ability to approximate non-linear rela-
horizon. Figure 1 presents the original photoelectric data
tionships [27]. They are a type of ANN capable of modeling
during daylight hours, from 8 am to 5 pm with an interval
complex functions. They are apt to ignore noise and irrelevant
of 5 minutes.
inputs, and they can easily adapt their weights. They are also
easy to use. The learning process of an MLP can be divided
B. PREDICTION TECHNIQUES FOR SOLAR POWER into two main phases: the input of data through the inputs
FORECASTING of the MLP and the correction of the errors, during which
Artificial Neural Networks (ANN) are part of the area of the errors are calculated by comparing the real data against
knowledge of Artificial Intelligence (AI) and Deep Learning, the answer that it delivers to the model through a technique
simulated through computer programs, to mimic the human known as Backpropagation. This iteration is repeated multi-
capacity to learn, memorize, and find relationships. ANNs, ple times to reduce the error, using an algorithm to obtain a
in particular, attempt to reproduce the behavior of biological result with better aptitude, being the Bayesian Regularization,
neural networks in an extremely simple way [23]. The ability one of the most common algorithms [28]. A multilayer per-
to learn nonlinear relationships and their ability to model ceptron is made up of an input layer, an output layer, and one
complex systems have made them a useful tool in different or more hidden layers; although it has been shown that for
scientific fields [24], [25]. The basic unit of ANNs is the arti- most problems, a single hidden layer will suffice [29], [30].
ficial neuron, which is a simplified mathematical abstraction In figure 2, a neuron j is represented, where xi are the inputs,
of the behavior of a biological neuron. ANNs are made up wij are the weights that relate each input i with the neuron j,
of a large number of artificial neurons grouped in layers and and yj is the output.

31694 VOLUME 10, 2022

M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques
The neuron performs two types of operations: first, the
propagation rule; and later, the activation function. The prop-
agation rule (Z ) is defined by the inputs and the synaptic
weights. The most commonly used is the sum of the product
of the inputs xi by the weights wij that join them to the neuron
j. This operation represents a linear function that passes by
way of the origin. To remove this limitation, a parameter
called threshold bj is added. This can be considered one more
input, with a fixed value of 1, whose weight bj must also be
updated. The propagation rule is calculated as
Xn
Zj = (xi .wij + bj ) (1)
i=1
where, Zj is the result of the propagation rule applied to
neuron j, xi is the input vector i, wij is the weight that joins
input i with neuron j, and bj is the threshold associated with
neuron j.
The activation function (A) is responsible for evaluating the
activation of the neuron and obtaining the output of neuron j.
It is determined based on the result of the propagation rule
such as


Aj = f Zj (2)
where, Aj is the activation of neuron j and f is the activation
function.
The most common activation functions are step, linear,
and sigmoidal functions, among which the logistic functions
and hyperbolic tangents stand out. Depending on the type of
problem to be solved, one type of response or another will be
required. For classification problems, where binary outputs
are desired, activation functions of the sigmoid type are often
used. These types of functions have a range of small values
with saturation at the extremes. The most used functions are
the logistic function, with a working range on the ordinate
from 0 to 1, and the hyperbolic tangent, with a working
range on the ordinate between −1 and 1. On the other hand,
to solve regression problems, they usually use functions of a
linear type, since more variability in the response is needed.
A neural network is organized into layers connected to each
other creating a neural network. A layer is understood to
be the set of neurons that are located at the same level in
the network and that process information at the same time.
In addition to the activation function of each neuron, the
behavior of the neural network depends on the topology and
the training carried out to establish the value of the weights.
D. LONG SHORT TERM MEMORY (LSTM)
Due to the rapid growth of the field of deep learning, which
has manifested itself in the development of new technologies
and architectures as well as greater integration in many areas
of research, structures such as long-term memory (LSTM)
have been recently used in the development of new solar fore-
casting techniques [31]. LSTM networks are currently one
of the most popular models in deep learning applied to time
series prediction. This kind of prediction is a difficult problem
due to the presence of long-term trends, seasonal and cyclical
VOLUME 10, 2022
fluctuations, and random noise [32]. LSTM structures are dis-
tinguished by their ability to model and recognize temporal
patterns from data sequences, with the presence of memory
cells and the way information is transferred between their
units. In this way, they are able to process the sequences and
records of the available operational data, thus extracting tem-
porary information that allows reducing forecast errors [9].
In general, past events influence future events, and based on
this idea, the recurrent neural network has a structure where
information from the previous step is passed to the next step
and used for estimation. Accordingly, the recurrent neural
network has achieved great results in estimating sequential
information, that is, time series data. However, as the length
of the past data, required for data estimation, increases, the
vanishing gradient problem occurs in the existing recurrent
neural network. This problem can be solved by using the gates
of the LSTM algorithm as proposed in [7], [33]. The networks
of Long Short-Term Memory (LSTM) are very similar to the
MLP structure. They have input layers, hidden layers, and
an output layer. However, LSTM in its hidden layer has a
memory unit [18]. LSTMs are an ANN type that is classified
as Recurrent Neural Networks, which are characterized by
not being strictly fed forward, such as MLPs. This is because
LSTMs use inputs from previous iterations for future output
calculations, thus providing feedback. This type of ANN is
potentially more powerful than MLP, and they are character-
ized by showing a temporary behavior [15]. The memory unit
consists of three gates: input gate, forget gate, and output gate
(Input gate (it ), Forget gate (ft ), and Output gate (ot )) and
a recurring connection as shown in Figure 3. The drive has
one input xt and two drive previous state feedbacks which are
the previous state output st−1 and the state variable ct−1 . The
gates use a sigmoid activation function g, while states use a
tanh function. The memory unit of the LSTM can be defined
by a set of equations, where w are the parameters and b are
the biases [18].
In general, the LSTM structure consists of three different
layers: forget, input and output. In the LSTM architecture,
first, xt and st−1 information are used as inputs, and it is
decided which information to delete. These operations are
done in the forget layer ft determined by


ft = g wxf xt + whf st−1 + bf (3)
where, g is the activation function, which is the sigmoid in
this work.
In the second step, the input layer it , where new informa-
tion will be determined, is given by
it = g (wxi xt + whi st−1 + bi ) (4)
Then, the candidate information, that will form the new infor-
mation, is expressed by
ct = ft .ct−1 + it .it t (5)
it t = tanh (wxc xt + whc st−1 + bi ti ) (6)
31695
 M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques

FIGURE 3. Diagram of operation of a cell of the long short-term memory

networks.

Finally, the output data is obtained by using the following

expressions in the output layer.
FIGURE 4. Shows one day 30 minutes ahead (December 31).
ot = g (wxo xt + who st−1 + b0 ) (7) TABLE 1. Performance evaluation for the winter day.
st = ot .tanh (ct ) (8)
The process described above continues iteratively. Weight
parameters (w) and bias parameters (b) are learned by the
model in a way that minimizes the difference between actual TABLE 2. Performance evaluation for the summer day.
training values and LSTM output values [34], [35].

E. PERFORMANCE EVALUATION
Four different statistical evaluation criteria were used to
evaluate the prediction performance of the proposed LSTM
model. These criteria are: IV. RESULTS AND DISCUSSIONS
1. Mean absolute error (MAE) In order to implement the proposed model, 43,800 records
It expresses the mean absolute deviation of the difference were taken during daylight hours with a 5-minute solar inter-
between the predicted values and the actual values and is val in Halifax, corresponding to the days from January 1,
calculated by 2017 to December 31, 2017 available at the Nova Scotia
Community College (NSCC) and to find the missing values.
1 XN A prediction was made every 30 minutes for a winter day
MAE = |oi − pi | (9)
N i=1 (December 31), and a summer day (June 30) for the total of
2. Root mean square error (RMSE) 7300 values, which were divided into 7280 training data, and
It represents the standard deviation of the estimation errors the last 20 for the prediction test. To evaluate the effectiveness
and is calculated by of the proposed method, MATLAB R2019b was used for
s
PN a LSTM training process with an initial learning rate of
2
i=1 (oi − pi ) 0.01 and a maximum epoch of 1000. A case study was car-
RMSE = (10)
N ried out using original photovoltaic data for one year during
daylight hours, from 8 am to 5 pm at 5-minute intervals.
3. Mean absolute percentage error (MAPE)
In this work, data is summarized at 30-minute intervals to
It measures the prediction accuracy of the models as a
make a half-hour forecast for the next day. In Figures 4 and 5,
percentage and is calculated by
  the actual graphs of daily power generation are shown in
1 XN  oi − pi  comparison with the graphs expected in the forecast models.
MAPE = (11)
N i=1  o 
i Figure 4 shows the results for one day in December, and
figure 5 shows the results for one day in June. The results
4. Coefficient of determination (R2 ) show that the forecast is close to the actual data, especially
It represents the strength of the linear relationship between during the winter period when the system did not generate
the predicted values of the models and the actual values and much power, and in the summer, the LSTM model has a
is calculated by significant advantage over the MLP model.
PN
(oi − pi )2 Tables 1 and 2 show the performance criteria for the two
2
R = 1 − Pi=1 N
(12) models, for the winter day and the summer day, respectively.
i=1 (oi − õ)
2
In general, when the results are examined, it has been shown
where, N is the number of samples used for statistical evalu- that the LSTM model gives the best results in all performance
ation criteria, oi is the actual value of the observation, and pi criteria on both days.
is the forecasted value of the observation and õ is the average The forecasted values of the proposed LSTM model used
of the actual observation values. to predict the solar power and the time graph showing the

31696 VOLUME 10, 2022

M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques
FIGURE 5. Shows one day 30 minutes ahead (June 30).
FIGURE 6. Prediction result of June (30 minutes interval).
FIGURE 7. Forecast results in June compared to actual values and RMSE.
actual values are shown in Figure 6. The actual 30 minute
ahead values for the June data test and the LSTM forecast
value are used in the chart. When the forecasted values of
the LSTM model are compared with the actual values, the
observed consistency in all training epochs is very close to
the real data; however, there are no very sharp fluctuations.
It is also reported that the forecasted and actual values show
the same trend. This means the LSTM algorithm rarely under-
goes gradient exploding or a vanishing gradient.
Figure 7 shows the results of the LSTM solar forecast
models for the month of June 2017. The graph represents
the daily average solar energy during daylight hours from
8 am to 5 pm. Although the graph shows that the prediction
is close to reality, solar energy is not only affected by the
parameters of this model. However, it differs due to various
VOLUME 10, 2022
factors of changes in the atmosphere, such as water vapor,
clouds, or pollution.
The results show the importance of the time series used
to train the models. The quality of the data has a significant
impact on the performance obtained by the forecasting model,
especially the outliers. A specific time series may perform
better with one model, while the same model may perform
poorly with another time series model. Consequently, it is
convenient to consider multiple models to search for suitable
predictions. However, there are several techniques for time
series forecasting that can be implemented according to the
methodology proposed in this study.
V. CONCLUSION
The LSTM model, which is based on a deep learning
approach, is proposed to predict daily solar power values.
The prediction was evaluated using solar power data from
Halifax, Nova Scotia, Canada over the course of a year
(January 1, 2017, to December 31, 2017). This data was
split into two categories: training and test sets. While only
the training data was used in the learning process of the
model, the test data was not used in the learning process.
The task of predicting photovoltaic power for the coming
days at 30-minute intervals was considered. The results of
the suggested model were compared using the Multi-Layer
Perceptron (MLP) algorithm, which is the most extensively
used technique in the literature, in order to evaluate and exam-
ine their correctness and performance. When compared to the
MLP algorithm, the prediction performance of the suggested
LSTM model offered more effective values in all performance
parameters MAE, MAPE, RMSE, and R2 for each category
of days, according to the findings of simulation tests and the
results presented in Tables 1 and 2. The proposed model gives
trustworthy data that will allow photovoltaic power plants
to operate more efficiently in the future. The combination
formed by the concepts of artificial intelligence and energy
efficiency appears to have a bright future, particularly in
terms of boosting energy sustainability, decarburization, and
electrical sector digitization. The importance of data process-
ing (time series) utilized to train models was highlighted in
this study. A specific time series may perform better with one
model, while the same model may have poorer performance
with another time series model. This reason suggests focusing
on research and development in multiple models in order to
arrive at predictions with high suitability. In addition, the
preprocessing approaches used to reduce noise, eliminate
outliers, and reduce errors from prediction models should be
taken into account. When used to predict solar energy, artifi-
cial intelligence algorithms have demonstrated their aptitude
and superiority in obtaining favorable outcomes. However,
obtaining such results would necessitate a significant amount
of hyperparameter adjustment. Furthermore, the quality of
the data, particularly the outliers, has a substantial impact
on the prediction model’s performance. Several time series
prediction approaches can be implemented using the method-
ology provided in this work.
31697
 M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques
REFERENCES
[1] D. Anderson and M. Leach, ‘‘Harvesting and redistributing renewable
energy: On the role of gas and electricity grids to overcome intermittency
through the generation and storage of hydrogen,’’ Energy Policy, vol. 32,
no. 14, pp. 1603–1614, Sep. 2004.
[2] T. Rojat, R. Puget, D. Filliat, J. Del Ser, R. Gelin, and N. Díaz-Rodríguez,
‘‘Explainable artificial intelligence (XAI) on TimeSeries data: A survey,’’
2021, arXiv:2104.00950.
[3] G. Briganti and O. Le Moine, ‘‘Artificial intelligence in medicine: Today
and tomorrow,’’ Frontiers Med., vol. 7, p. 27, Feb. 2020.
[4] J. Bedi and D. Toshniwal, ‘‘Deep learning framework to forecast electricity
demand,’’ Appl. Energy, vol. 238, pp. 1312–1326, Mar. 2019.
[5] L. Alzubaidi, J. Zhang, A. J. Humaidi, A. Al-Dujaili, Y. Duan,
O. Al-Shamma, J. Santamaría, M. A. Fadhel, M. Al-Amidie, and
L. Farhan, ‘‘Review of deep learning: Concepts, CNN architectures,
challenges, applications, future directions,’’ J. Big Data, vol. 8, no. 1,
pp. 1–74, Dec. 2021.
[6] S. Hochreiter, ‘‘The vanishing gradient problem during learning recurrent
neural nets and problem solutions,’’ Int. J. Uncertainty, Fuzziness Knowl.-
Based Syst., vol. 6, no. 2, pp. 107–116, Apr. 1998.
[7] S. Hochreiter and J. Schmidhuber, ‘‘Long short-term memory,’’ Neu-
ral Comput., vol. 9, no. 8, pp. 1735–1780, 1997. [Online]. Available:
https://deeplearning.cs.cmu.edu/pdfs/Hochreiter97-lstm.pdf
[8] S. A. Kalogirou, ‘‘Artificial neural networks in renewable energy sys-
tems applications: A review,’’ Renew. Sustain. Energy Rev., vol. 5, no. 4,
pp. 373–401, Dec. 2001.
[9] M. Elsaraiti and A. Merabet, ‘‘A comparative analysis of the ARIMA
and LSTM predictive models and their effectiveness for predicting
wind speed,’’ Energies, vol. 14, no. 20, p. 6782, Oct. 2021, doi:
10.3390/en14206782.
[10] S. R. Young, D. C. Rose, T. P. Karnowski, S.-H. Lim, and R. M. Patton,
‘‘Optimizing deep learning hyper-parameters through an evolutionary
algorithm,’’ in Proc. Workshop Mach. Learn. High-Perform. Comput. Env-
iron., Nov. 2015, pp. 1–5.
[11] M. Q. Raza and A. Khosravi, ‘‘A review on artificial intelligence based
load demand forecasting techniques for smart grid and buildings,’’ Renew.
Sustain. Energy Rev., vol. 50, pp. 1352–1372, Oct. 2015.
[12] S. A. Kalogirou, ‘‘Solar thermal power systems,’’ in Solar Energy Engi-
neering, 2009.
[13] A. Azadeh, R. Babazadeh, and S. M. Asadzadeh, ‘‘Optimum estimation
and forecasting of renewable energy consumption by artificial neural
networks,’’ Renew. Sustain. Energy Rev., vol. 27, pp. 605–612, Nov. 2013.
[14] M. Elsaraiti and A. Merabet, ‘‘Application of long-short-term-memory
recurrent neural networks to forecast wind speed,’’ Appl. Sci., vol. 11, no. 5,
p. 2387, Mar. 2021.
[15] J. Bottieau, F. Vallee, Z. De Greve, and J.-F. Toubeau, ‘‘Leveraging provi-
sion of frequency regulation services from wind generation by improving
day-ahead predictions using LSTM neural networks,’’ in Proc. IEEE Int.
Energy Conf. (ENERGYCON), Limassol, Cyprus, Jun. 2018, pp. 1–6.
[16] A. Rabehi, M. Guermoui, and D. Lalmi, ‘‘Hybrid models for global solar
radiation prediction: A case study,’’ Int. J. Ambient Energy, vol. 41, no. 1,
pp. 31–40, Jan. 2020, doi: 10.1080/01430750.2018.1443498.
[17] E. Pasero, G. Raimondo, and S. Ruffa, ‘‘MULP: A multi-layer perceptron
application to long-term, out-of-sample time series prediction,’’ in Proc.
Int. Symp. Neural Netw., Jun. 2010, pp. 566–575.
[18] Q. Xiaoyun, K. Xiaoning, Z. Chao, J. Shuai, and M. Xiuda, ‘‘Short-
term prediction of wind power based on deep long short-term memory,’’
in Proc. IEEE PES Asia–Pacific Power Energy Eng. Conf., Oct. 2016,
pp. 1148–1152.
[19] M. Abdel-Nasser and K. Mahmoud, ‘‘Accurate photovoltaic power fore-
casting models using deep LSTM-RNN,’’ Neural Comput. Appl., vol. 31,
no. 7, pp. 2727–2740, Jul. 2019.
[20] S. Han, Y.-H. Qiao, J. Yan, Y.-Q. Liu, L. Li, and Z. Wang, ‘‘Mid-to-long
term wind and photovoltaic power generation prediction based on copula
function and long short term memory network,’’ Appl. Energy, vol. 239,
pp. 181–191, Apr. 2019.
[21] M. Diagne, M. David, P. Lauret, J. Boland, and N. Schmutz, ‘‘Review
of solar irradiance forecasting methods and a proposition for small-scale
insular grids,’’ Renew. Sustain. Energy Rev., vol. 27, pp. 65–76, Nov. 2013.
[22] A. Mellit and A. M. Pavan, ‘‘A 24-h forecast of solar irradiance using
artificial neural network: Application for performance prediction of a
grid-connected PV plant at Trieste, Italy,’’ Sol. Energy, vol. 84, no. 5,
pp. 807–821, May 2010.
31698
[23] M. Hagan, M. Beale, and H. Demuth, Neural Network Design. Boston,
MA, USA: PWS Publishing Company, 1996.
[24] A. K. Yadav and S. S. Chandel, ‘‘Solar radiation prediction using artifi-
cial neural network techniques: A review,’’ Renew. Sustain. Energy Rev.,
vol. 33, pp. 772–781, May 2014.
[25] C. Paoli, C. Voyant, M. Muselli, and M.-L. Nivet, ‘‘Forecasting of pre-
processed daily solar radiation time series using neural networks,’’ Sol.
Energy, vol. 84, no. 12, pp. 2146–2160, Dec. 2010.
[26] F. S. Tymvios, S. C. Michaelides, and C. S. Skouteli, ‘‘Estimation of
surface solar radiation with artificial neural networks,’’ in Modeling
Solar Radiation at the Earth’s Surface. Berlin, Germany: Springer, 2008,
pp. 221–256.
[27] R. Reed and R. J. Marks, Neural Smithing: Supervised Learning in Feed-
forward Artificial Neural Networks. Cambridge, MA, USA: MIT Press,
Feb. 1999.
[28] R. Velo, P. López, and F. Maseda, ‘‘Wind speed estimation using multilayer
perceptron,’’ Energy Convers. Manage., vol. 81, pp. 1–9, May 2014.
[29] K.-I. Funahashi, ‘‘On the approximate realization of continuous mappings
by neural networks,’’ Neural Netw., vol. 2, no. 3, pp. 183–192, Jan. 1989.
[30] J. H. Byrne, ‘‘Introduction to neurons and neuronal networks,’’ in Text-
book for the Neurosciences, 2013. [Online]. Available: https://nba.uth.
tmc.edu/neuroscience/s1/introduction
[31] S. Srivastava and S. Lessmann, ‘‘A comparative study of LSTM neural net-
works in forecasting day-ahead global horizontal irradiance with satellite
data,’’ Sol. Energy, vol. 162, pp. 232–247, Mar. 2018.
[32] S. Muzaffar and A. Afshari, ‘‘Short-term load forecasts using LSTM
networks,’’ Energy Proc., vol. 158, pp. 2922–2927, Feb. 2019.
[33] F. A. Gers, N. N. Schraudolph, and J. Schmidhuber, ‘‘Learning precise
timing with LSTM recurrent networks,’’ J. Mach. Learn. Res., vol. 3, no. 1,
pp. 115–143, 2003.
[34] A. Sagheer and M. Kotb, ‘‘Time series forecasting of petroleum produc-
tion using deep LSTM recurrent networks,’’ Neurocomputing, vol. 323,
pp. 203–213, Jan. 2019.
[35] T. Fischer and C. Krauss, ‘‘Deep learning with long short-term memory
networks for financial market predictions,’’ Eur. J. Oper. Res., vol. 270,
no. 2, pp. 654–669, Oct. 2018.
MEFTAH ELSARAITI (Member, IEEE) received
the B.S. degree in physics from Tripoli Uni-
versity, Tripoli, Libya, in 1984, and the M.S.
degree in mechanical engineering from Bradford
University, Bradford, U.K., in 2005. He is cur-
rently pursuing the Ph.D. degree in applied science
with Saint Mary’s University, Halifax, Canada.
From 1985 to 1990, he was a Research Assistant
with the Iron and Steel Training Center, Misurata,
Libya, and from 2005 to 2013, he was a Lecturer
with the Higher Institute for Technical Sciences Misurata, Libya. His cur-
rent research interest includes developing models for predicting renewable
energies.
ADEL MERABET (Senior Member, IEEE)
received the Ph.D. degree in engineering from
the University of Québec at Chicoutimi, Canada,
in 2007. He joined Saint Mary’s University,
Halifax, Canada, in 2009. He is currently an Asso-
ciate Professor with the Division of Engineering.
From 2016 to 2017, he was a Visiting Academic
with the Department of Sustainable and Renew-
able Energy Engineering, University of Sharjah,
Sharjah, United Arab Emirates. In 2019, he was a
recipient of the Tan Chin Tuan Exchange Fellowship in Engineering with
Nanyang Technological University in Singapore. In 2021, he was a Visiting
Research Scientist with the Advanced Power and Energy Center, EECS
Department, Khalifa University, Abu Dhabi, United Arab Emirates. His
research interests include renewable (wind-solar) energy conversion systems,
energy management, advanced control, electric drives, artificial intelligence,
and smart grid.
VOLUME 10, 2022